Mario Rosario Guarracino

A classification method based on generalized eigenvalue problems

Binary classification refers to supervised techniques that split a set of points in two classes, with respect to a training set of points whose membership is known for each class. Binary classification plays a central role in the solution of many scientific, financial, engineering, medical and biological problems. Many methods with good classification accuracy are currently available. This work shows how a binary classification problem can be expressed in terms of a generalized eigenvalue problem. A new regularization technique is proposed, which gives results that are comparable to other techniques in use, in terms of classification accuracy. The advantage of this method relies in its lower computational complexity with respect to the existing techniques based on generalized eigenvalue problems. Finally, the method is compared with other methods using benchmark data sets.

Cold adaptation shapes the robustness of metabolic networks in Drosophila melanogaster

When ectotherms are exposed to low temperatures, they enter a cold‐induced coma (chill coma) that prevents resource acquisition, mating, oviposition, and escape from predation. There is substantial variation in time taken to recover from chill coma both within and among species, and this variation is correlated with habitat temperatures such that insects from cold environments recover more quickly. This suggests an adaptive response, but the mechanisms underlying variation in recovery times are unknown, making it difficult to decisively test adaptive hypotheses. We use replicated lines of Drosophila melanogaster selected in the laboratory for fast (hardy) or slow (susceptible) chill‐coma recovery times to investigate modifications to metabolic profiles associated with cold adaptation. We measured metabolite concentrations of flies before, during, and after cold exposure using nuclear magnetic resonance (NMR) spectroscopy to test the hypotheses that hardy flies maintain metabolic homeostasis better during cold exposure and recovery, and that their metabolic networks are more robust to cold‐induced perturbations. The metabolites of cold‐hardy flies were less cold responsive and their metabolic networks during cold exposure were more robust, supporting our hypotheses. Metabolites involved in membrane lipid synthesis, tryptophan metabolism, oxidative stress, energy balance, and proline metabolism were altered by selection on cold tolerance. We discuss the potential significance of these alterations.

Non-small cell lung cancer evaluated with quantitative contrast-enhanced CT and PET-CT: Net enhancement and standardized uptake values are related to tumour size and histology

Background Personalized cancer therapy remains a challenge. In this context, we attempted to identify correlations between tumour angiogenesis, tumour metabolism and tumour cell type. To this aim, we used single=phase multidetector computed tomography (MDCT) and hybrid positron emission tomography-computed tomography (PET/CT) to determine whether net enhancement and standardized uptake value (SUVmax) were correlated with tumour size and cytology in patients affected by non-small cell lung cancer (NSCLC). Material/Methods Our study included 38 patients (30 men, 8 women, mean age 70) with a NSCLC measuring between 3 cm and 7 cm, using a 16-slice multidetector CT (Brilliance Philips) and with PET-CT (Biograph 16 Siemens Medical Solutions). The following lesion parameters were evaluated: maximum diameter, medium density before contrast injection (CTpre), medium density after contrast injection (CTpost average), density in the most enhanced part of the lesion after contrast (CTpost max), net enhancement, SUVmax, age, and cytology. Correlation coefficient and p-value were computed for each pair of variables. In addition, correlations were computed for each pair of variables, and for all combinations of tumour types. We focused on subsets of data with more than 10 observations, and with correlation r>0.500 and p<0.05. Results A weak correlation (r=0.32; p=0.048) was found between SUVmax and tumour size; the correlation was stronger for masses larger than 31 mm (r=0.4515; p=0.0268). No other correlations were found among the variables examined. Conclusions Our data may have prognostic significance, and could lead to more appropriate surgical treatment and better treatment outcome.

Incremental Classification with Generalized Eigenvalues

Supervised learning techniques are widely accepted methods to analyze data for scientific and real world problems. Most of these problems require fast and continuous acquisition of data, which are to be used in training the learning system. Therefore, maintaining such systems updated may become cumbersome. Various techniques have been devised in the field of machine learning to solve this problem. In this study, we propose an algorithm to reduce the training data to a substantially small subset of the original training data to train a generalized eigenvalue classifier. The proposed method provides a constructive way to understand the influence of new training data on an existing classification function. We show through numerical experiments that this technique prevents the overfitting problem of the earlier generalized eigenvalue classifiers, while promising a comparable performance in classification with respect to the state-of-the-art classification methods.

Robust generalized eigenvalue classifier with ellipsoidal uncertainty

Uncertainty is a concept associated with data acquisition and analysis, usually appearing in the form of noise or measure error, often due to some technological constraint. In supervised learning, uncertainty affects classification accuracy and yields low quality solutions. For this reason, it is essential to develop machine learning algorithms able to handle efficiently data with imprecision. In this paper we study this problem from a robust optimization perspective. We consider a supervised learning algorithm based on generalized eigenvalues and we provide a robust counterpart formulation and solution in case of ellipsoidal uncertainty sets. We demonstrate the performance of the proposed robust scheme on artificial and benchmark datasets from University of California Irvine (UCI) machine learning repository and we compare results against a robust implementation of Support Vector Machines.

MedIGrid: a medical imaging application for computational Grids

In the last decades, diagnosing medical images has heavily relied on digital imaging. As a consequence, huge amounts of data produced by modern medical instruments need to be processed, organized, and visualized in a suitable response time. Many efforts have been devoted to the development of digital Picture Archiving and Communications Systems (PACS) which archive and distribute image information across a hospital and provide Web access to avoid the expensive deployment of a large number of such systems. On the other hand, this approach does not solve problems related to the increasing demand of high performance computing and storage facilities, which cannot be placed within a hospital. In this work we describe MedIGrid, an application that enables nuclear doctors to transparently use high performance computers and storage systems for the PET/SPECT (Positron Emission Tomography/Single Photon Emission Computed Tomography) image processing, management, visualization and analysis. MedIGrid is the result of the joint efforts of a group of researchers committed to the development of a distributed application to test and deploy new reconstruction methods in clinical environments. The outcomes of this work include a set of platform independent software tools to read medical images, control the execution of computing intensive tomographic algorithms, and explore the reconstructed tomographic volumes. In the paper we describe how the collaboration among different research groups has contributed to the integration of the application into a single framework. The results of our work are discussed.

Multiclass Generalized Eigenvalue Proximal Support Vector Machines

Support Vector Machines represent state of the art in supervised learning. Recently, the Regularized Generalized Eigenvalue Classifier (ReGEC) extension has been proposed to solve binary classification problems. In the present work we describe MultiReGEC, a novel technique that generalizes ReGEC to multiclass classification problems. This method is based on statistical and geometrical considerations, providing strong fundamentals to the proposed extension. After a detailed description of the MultiReGEC algorithm, we show, through extensive numerical experiments, that the accuracy of the proposed algorithm well compares with other de facto standard techniques.

Transcriptator: An Automated Computational Pipeline to Annotate Assembled Reads and Identify Non Coding RNA

RNA-seq is a new tool to measure RNA transcript counts, using high-throughput sequencing at an extraordinary accuracy. It provides quantitative means to explore the transcriptome of an organism of interest. However, interpreting this extremely large data into biological knowledge is a problem, and biologist-friendly tools are lacking. In our lab, we developed Transcriptator, a web application based on a computational Python pipeline with a user-friendly Java interface. This pipeline uses the web services available for BLAST (Basis Local Search Alignment Tool), QuickGO and DAVID (Database for Annotation, Visualization and Integrated Discovery) tools. It offers a report on statistical analysis of functional and Gene Ontology (GO) annotation’s enrichment. It helps users to identify enriched biological themes, particularly GO terms, pathways, domains, gene/proteins features and protein—protein interactions related informations. It clusters the transcripts based on functional annotations and generates a tabular report for functional and gene ontology annotations for each submitted transcript to the web server. The implementation of QuickGo web-services in our pipeline enable the users to carry out GO-Slim analysis, whereas the integration of PORTRAIT (Prediction of transcriptomic non coding RNA (ncRNA) by ab initio methods) helps to identify the non coding RNAs and their regulatory role in transcriptome. In summary, Transcriptator is a useful software for both NGS and array data. It helps the users to characterize the de-novo assembled reads, obtained from NGS experiments for non-referenced organisms, while it also performs the functional enrichment analysis of differentially expressed transcripts/genes for both RNA-seq and micro-array experiments. It generates easy to read tables and interactive charts for better understanding of the data. The pipeline is modular in nature, and provides an opportunity to add new plugins in the future. Web application is freely available at: http://www-labgtp.na.icar.cnr.it/Transcriptator

Application oriented brokering in medical imaging: algorithms and software architecture

This paper describes algorithms and software architecture of a resource broker designed in the context of MedIGrid, a medical imaging application for the management, visualization and reconstruction on grids of medical images produced by PET/SPECT medical instruments. The broker allows the discovery and selection of suitable clusters of workstations for the execution of parallel image reconstruction algorithms. The proposed algorithms and software architecture are general with respect to possible application domains and are potentially useful in different grid environments.

Data mining in psychiatric research.

Mathematical sciences and computational methods have found new applications in fields like medicine over the last few decades. Modern data acquisition and data analysis protocols have been of great assistance to medical researchers and clinical scientists. Especially in psychiatry, technology and science have made new computational methods available to assist the development of predictive modeling and to identify diseases more accurately. Data mining (or knowledge discovery) aims to extract information from large datasets and solve challenging tasks, like patient assessment, early mental disease diagnosis, and drug efficacy assessment. Accurate and fast data analysis methods are very important, especially when dealing with severe psychiatric diseases like schizophrenia. In this paper, we focus on computational methods related to data analysis and more specifically to data mining. Then, we discuss some related research in the field of psychiatry.